Size Classification of Fine Particles by Opposing Jets - American

Jun 21, 1977 - (19) Box, G.E.P., Jenkens, G. M., “Time Series Analysis: Forecasting and Control”, Holden-Day, 1976. (20) Huber, P. J., Ann. Math. ...
2 downloads 9 Views 506KB Size
Conf., Research Triangle Park, N.C., 1976. (13) Westberg, H., Allwine, K. J., Elias, D., “Ozone/Oxidants-Interactions with the Total Environment”, Air Pollution Control Assoc., Pittsburgh, Pa., 1976. (14) Wolff, G., Stasiuk,W., Coffey, P., Pasceri,R., paper 75-58.6,Air Pollution Control Assoc. Annual Meeting, Boston, Mass., 1975. (15) Tiao, G. C., Phadke, M. S., Box, G.E.P.,J.Air Pollut. Control Assoc., 26,485 (1976). (16) Elkus. B. E.. Wilson. K. R.. Atmos. Enuiron.. in Dress. (17) Wilk, M.B.,’Gnanadesikan,R., Biometrika, 55,l (1968). (18) Gifford,F. A., Hanna, S. R., Atmos. Enuiron., 7,131 (1973). ~~

~

~

I

~~~

(19) Box, G.E.P.,Jenkens, G. M., “Time Series Analysis: Forecasting and Control”,Holden-Day,1976. (20) Huber, P. J., Ann. Math. Stat., 43,1041 (1976). (21) Bach, W. D., Decker, C. E., Hamilton, H. L., Matus, L. K., Ripperton, L. A., Royal, T. M., Worth, J.J.B., “Investigation of High Ozone Concentration in the Vicinity of Garnett County, West Virginia”, Rep. of Project 410-764, Research Triangle Institute, Research Triangle Park, N.C., 1973.

r - - - -

Received fer review June 21,1977. Accepted November 3,1977.

Size Classification of Fine Particles by Opposing Jets Klaus Willeke” and Robert E. Pavlik’ Kettering Laboratory, University of Cincinnati, Cincinnati, Ohio 45267

A new technique for the size classification of airborne particles is developed by directing an aerosol-laden air flow as a jet against an axially opposing clean air jet of equal strength. The large, high-inertia particles are transferred into the clean airstream, while the smaller, low-inertia particles remain with the original airstream. The technique, tested for particles in the 1-5-pm size range, gives a sharp size classification with a geometric standard deviation ranging from 1.04 to 1.15 and does not have the particle bounce and reentrainment problem associated with conventional impaction techniques. Size classification measurements with a micrometer-positioned needle feed of aerosols into the upper air jet showed that the separation efficiency varies somewhat with radial position of the particles in the airjet. The objective of this study was to develop a new method of inertial particle-size classification applicable to the measurement of particle-size distributions in ambient or industrial environments and to the size segregation of polydisperse powders for industrial applications or toxicological studies requiring narrow size fractions. In this technique, schematically represented in Figure 1, particles of high inertia are transferred out of a particle-laden air jet into an equally strong, but axisymmetrically opposed, clean air jet, while smaller particles are retained in the original airstream. The particles in the two separate, size-fractionated airstreams may then be collected and chemically analyzed by ducting the air streams through the filter or sensor that is most suitable for the intended use. This technique thus avoids the uncertainty of bounced or reentrained particles ( I , 2) common to cyclones and conventional solid-plate impactors. The dynamics of solid plate impaction and the resulting particle-size classification characteristics have been studied extensively (3-6) and are reasonably well understood because most of the flow field in the impaction region is laminar so that the streamlines and the particle trajectories within the flow field are predictable and scalable. In the virtual or dichotomous impactor (7) the air turbulence created by the impingement of the aerosol-laden air jet into a quiescent air space interacts with the original airstream. The air turbulence makes the genera!ized study of the interaction phenomenon difficult, if not impossible. In practice, the air turbulence is removed or significantly diminished by pumping the void at about one-eighth of the total flow rate (7). This extraction flow may, however, contain some un-

Present address, US.Navy Regional Medical Center, San Diego, Calif. 92134. 0013-936X/78/0912-0563$01 .OO/O

LOW-INERTIA

HIGH-INERTIA...’’

Figure 1. Schematic diagram of opposing-jet particle-size classifier

Dimensions: W=3.2mm, T =

13.7mm,S= 1.5mm,P=O.l5mm,H=4mm,

N = 1.6 mm, C = 6.7 mm, inner diameter of collecting chamber = 47.4 mm

classified particles in addition to the classified, high-inertia particles. In the opposing-jet classifier described here, the high-inertia particles cross a fluid interface, and the flow in the interaction region is essentially laminar, because the pressure gradient of the aerosol-laden airjet is countered by an equally strong pressure gradient of the opposing jet. Previous attempts in using opposing jets for particle-size classification did not result in sharp separation efficiency curves (8-10).

Experimental Opposing Jet Classifier. The opposing jet classifier (see Figure 1)was designed to operate a t the flow rate of the particle sensor used in the experiment and was to give a 50% cut size in the particle-size range of 2-3.5 pm. Moderate geometric scaling of the device can then shift the cut point to the respirable cut size ( 1 2 ) of 3.5 pm or to the physical cut size of about 2 pm. The latter cut size has been observed to give a minimum in the multimodal mass distribution found in most ambient

0 1978 American Chemical Society

Volume 12,Number 5,May 1978 563

air environments (12). The two air flows have volumetric flow rates of 118 cm3/s (0.25 cfm) each and are drawn through 60 degree tapered inlets into round nozzles of inside width W = 3.2 mm (l/8 in.). The Reynolds number of the air jets is 3140. The separation distance from the nozzle exit plane to the fluid interface between the two opposing jets corresponds to about half the nozzle width ( 5 ) .The inner diameter of the collecting chamber is sufficiently large to decelerate the air flow that radially emanates from the impaction region. Particle-Size Classification and Loss Analysis. The experimental setup is schematically represented in Figure 2. The test aerosols were generated by a vibrating-orifice monodisperse-aerosol generator (Thermo-Systems, Inc., St. Paul, Minn.). After start-up of the generator, the pressure in the syringe-fed liquid supply line was quite high and decreased very slowly. The pressure was therefore monitored, and the liquid outflow was fed through a needle valve opened after start-up until the pressure was reduced to the stable equilibrium value (13). The aerosols were discharged to near-Boltzmann equilibrium by a lO-mCi, Kr-85 radioactive source. When tests were performed for the maximum possible particle loss that may occur in the system, liquid aerosols were generated from solutions of oleic acid and isopropyl alcohol (5).When testing for minimum particle loss with a solid particle, potassium biphthalate, KHCsH404, in solution with water was used ( 1 4 ) . In the latter case, dryness of the aerosols and of the air itself was ensured by passing the aerosol flow, QA or Qk, through a cylindrical diffusion dryer in which the aerosol stream is separated from desiccant silica gel by a fine-meshed screen. Additional particle loss analyses were performed by tagging the aerosols with uranine fluorescent dye and analyzing the aerosol deposits by spectrofluorometry (5). The opposing-jet classifier was operated inside a glove box continuously supplied with clean air and maintained a t a slightly positive pressure. The particle concentrations in the two flows emanating from the classifier were analyzed by a Climet Model CI-201 optical single particle counter (OPC) coupled to a multichannel analyzer (MCA). In the overall feed experiment, schematically represented on the left side of Figure 2, the aerosol flow Q A was diluted to a concentration suitable for optical particle counting by mixing it with the clean upper flow QCU.The size-separated, upper and lower flows, Qsv and QSL, were alternately coupled to the OPC. The reference aerosol concentration for the loss analysis was measured by passing the aerosol flow directly to the OPC which was operated a t the same flow rate. Radial Scan Measurements. Ideally, the high-inertia particles transfer from the upper air flow to the lower one, while the low-inertia particles remain in the upper air flow. In the nonideal case, the transferred particles may bounce back into the upper air jet. If the hole in the collecting plate is large, air turbulence will degrade the size classification, and edge-tone effects (15)may transfer particles to either one of the collecting chambers in an oscillating fashion. T o study these effects and their relationship to the radial position of the aerosol in the original airstream, a separate radial scan system (Figure 2) was designed. The aerosols were isokinetically fed into the clean, laminar airflow within the nozzle inlet through a hypodermic needle that was radially translated by a micrometer screw. The extremely small aerosol flow was 1/300th of the total flow and was ducted from the aerosol generator into the classifier in a straight, vertical line to minimize losses. To get the reference count for the loss and classification analyses, another chamber was built which bypassed the classifier and fed directly into the OPC (see Figure 2). The pressure and flow rates in this chamber were made equal to those of the micrometer-feed chamber by making the 564

Environmental Science & Technology

pressure differentials equal to each other in the two systems. Bubble meter measurements confirmed the equalization of the flow rates.

Results and Discussion Overall Feed. The separation efficiency is obtained by dividing the particle count in the lower outlet by the sum of the counts from the upper and lower outlets. As seen from Figure 3, the size-classification efficiency obtained with liquid oleic acid aerosols is very sharp. Assuming that the 16 and 84% efficiencies give a representative measure of spread ( 5 , 6 ) ,the geometric standard deviation, ug, for the size-classification curve is 1.04. The difference in count between classifier inlet and outlet, divided by the inlet count, gives the particle loss shown in the same figure. As seen, the losses for liquid aerosols are quite OMRpSL FEE0

RADIAL SCAN LOSS ANALYSIS

MICROMETER FEED

GLOVE BOX

MW FRINTER

OPC

Figure 2. Experimental setup Loss analysis device (in center) gave reference count during radial scan experiments. Devices to left and right differ by their inlet design. During overall feed experiments only left device mounted inside glove box: during radial scan experiments, other two devices mounted inside. VOMAG = vibrating orifice monodisperse aerosol generator, OPC = optical particle counter, MCA = multichannel analyzer

TEST AEROSOL: LlQUiD OLEIC ACID

a

2

60

z \

AERODYNAMIC EQUIVALENT DIAMETER, DAE,p m

Figure 3. Performance of opposing-jet classifier, measured with liquid aerosols

' A U T TCENTERUNE a

100

-

90

80

3

if

2

70

;

% 60 TEST LlOUlDAEROSOL OLEIC ACID

U

50

w

DIE * 2 I6pm

0 0

I

3

2

4

5

6

7

8

D

9

II

12

FEED DISTANCE FROM CENTER, m m

loo

I o.

90

0-

s

-

I

801

I

I

I

I

I

P"

TEST AEROSOL SOLID POTASSIUM BIPHTHALATE

b

,---CENTERUNE ( (

9o

(

(

waLL7 (

1

(

~

,

~

TEST AEROSOL. UQUlD OLEIC ACID

-

0

I

1 , 1

40

P 30

PO

40 10

LL W LL

0

o

'

I

1

2

1

3

1

4

1

5

1

6

1

7

i e

1

9

'

IO

1

II

12

F E D DISTANCE FROM C E M E R , m m

Figure 6. Radial scan for liquid aerosols a. Separation efficiency

2

3

4

5 AERODYNAMIC EQUIVALENT DIAMETER, DAE,&m Figure 4. Performance measured with solid aerosols

%lid Potassium Bbphtholote Aerosol

L.-

I

I -

OpposingExperiment0 O,e,c JetI, Clossifler,

Acid Aerosol

1

.d ;I

n'4 16 0 ! 2 y 3

fication compared to conventional solid-plate impactors which generally use several nozzles for a given impactor stage to reduce bounce and reentrainment of particles from the impact surfaces. Tests in ambient air environments containing sticky and nonsticky aerosols would result in performance curves within the limits defined by the presented tests with liquid and solid aerosols. Reduction in particle losses through improvements in design of the opposing-jet classifier is expected to result in size-classification curves closer to the solid than the liquid aerosol case (Figure 4). Observations of the deposited fluorescent-dye-tagged liquid aerosols using ultraviolet light showed that the aerosols not only impact onto the inside rim of the separation plate hole but also onto both sides of the separation plate in rings reminiscent of the halo pattern observed when solid particles are impacted onto a solid plate (16). Liquid particles impacted into a fluid interface might behave similar to solid particles bouncing into a solid plate. The separation efficiency curves of Figures 3 and 4 are replotted in Figure 5 against the square root of Stokes number (61, which is a nondimensional representation of particle size. They are compared to the numerically calculated efficiency curves for solid-plate impaction ( 3 , 4 )with similar inlet design. Experiments with solid-plate impactors have shown good agreement between theory and experiment (5,6).One of the boundary conditions in solid-plate impaction is that the radial air velocity on the impaction plate is zero, whereas such a radial component will exist in a fluid interface away from the stagnation point. The radial air velocity at the interface should

jL,qu,d 1

Opposing- Jet Classifier Experlmentol,

'Ll

b. Particle loss

0

7 015

016

017

0!8

m Figure 5. Nondimenslonal comparison of performance data

Volume 12, Number 5, May 1978 565

~

make the particle transfer across the fluid interface less efficient, which is borne out by the displacement of the opposing-jet classification curve to a somewhat higher particle size. It appears from Figure 5 that the numerically calculated curves for solid-plate impaction may be used to estimate the performance of the opposing-jet classifier. Numerical calculations performed for the opposing-jet geometry will provide better predictions. Radial Scan. The particle separation efficiencies and losses for radial scans using four different sizes of liquid oleic acid aerosols are shown in Figure 6. Traverses. from wall-to-wall showed that a slight misalignment in the two opposing jets or in the separation plate resulted in somewhat different performance curves to either side of the center line. To facilitate interpretation, the data for equal radial positions on either side of the centerline are averaged in Figure 6. The radial scan data show some variations of size classification and particle loss as a function of radial position in the original airstream. It appears that particle losses are less for aerosols coming from the centerline because they can best clear the separation plate. These centerline particles also have a slightly higher than average size separation efficiency, which may be due to a somewhat higher than average velocity in the core of the nozzle. The performance near the wall also deviates from the average which may, in part, be caused by an interference of the needle feed with the wall. However, when all radial data are weighted relative to the area they represent, the calculated overall size classification and particle loss agrees closely with the data obtained through overall feed experiments, as seen in Figure 3. During initial testing the velocity of the needle feed was about 10 times that of the average air velocity and resulted in large radial variation of the performance curves. Although the needle flow changed the main flow, one may conclude from these initial observations that there appears to be a dependence of performance on radial position. Further tests with thinner plates and with different-sized separation holes will be performed and are likely to provide further information on these relationships. In conclusion, this new technique gives a sharp particle-size classification. Future work will concentrate on the reduction

of particle losses. Assuming that a minimum separation plate thickness is needed for mechanical strength, a device with a higher flow rate will have a higher ratio of deflected flow height to separation plate thickness, and should, therefore, have lower losses. The deflected flow entrains air in the collecting chamber, thus giving rise to a secondary flow pattern that may deposit some particles onto the surrounding walls and onto the separation plate itself. This phenomenon, common to all types of impaction devices, can generally be reduced by designing the collecting chamber to be relatively large.

Literature Cited (1) Dzubay, T. G., Hines, L. E., Atmos. Enuiron., 9,l-6 (1975). (2) Rao, A. K., Whitby, K. T., Am. Ind. Hyg. Assoc. J., 38, 174-9 (1977). (3) Marple, V. A., Liu, B.Y.H., Environ. Sci. Technol., 8, 643-54 (1974). (4) Marple, V. A., Willeke, K., Atmos. Enuiron., 10,891-6 (1976). (5) Willeke, K., Am. Ind. Hyg. Assoc. J., 36,683-91 (1975). (6) Willeke, K., McFeters, J. J., J. Colloid Interface Sci., 53,121-7 (1975). (7) Dzubay, T. G., Stevens, R. K., Enuiron. Sci. Technol., 9,663-8 (1975). ( 8 ) Luna, R. E., PhD thesis, Princeton University, Princeton, N.J., 1965. (9) Hall, R. E., MS thesis, University of Kentucky, Lexington, Ky., 1970. (10) Mears, C. E., MS thesis, University of Kentucky, Lexington, Ky., 1973. (11) Mercer, T. T., “Aerosol Technology in Hazard Evaluation”, Academic Press, New York, N.Y., 1973. (12) Willeke, K., Whitby, K. T., J. Air Pollut. Control Assoc., 25, 529-34 (1975). (13) Baron, P., NIOSH, Cincinnati, Ohio, private communication, 1977. (14) Reischl, G., John, W., Devor, W., J . Aerosol Sci., 8, 55-65 (1977). (15) Karamcheti, K., Bauer, A. B., SUDAER No. 162, Stanford University, Stanford, Calif., 1963. (16) May, K. R., J. Aerosol Sci., 6,403-11 (1975). Received for review September 9,1977. Accepted November 7,1977. Research was supported by Grant ENG 77-04667from the National Science Foundation and was part of a Center Program supported by Grant ES-00159-11from the National Institute of Enuironmental Health Sciences. Support of one of the authors (R.E.P.)by a PhD Outseruice Training Program, Bureau of Medicine and Surgery, U.S. Nauy.

Composition and Size Distributions of Particles Released in Refuse Incineration Robert R. Greenberg’, William H. Zoller, and Glen E. Gordon’ Department of Chemistry, University of Maryland, College Park, Md. 20742

In recent years there has been increasing concern about toxic elements in urban atmospheres. Some chemical forms of the following elements are generally considered to be toxic to humans when deposited in the lungs: Be, Cr, Ni, As, Se, Cd, Sn, Sb, Hg, and P b ( I , 2). Before optimum control strategies for toxic species can be devised, major sources of the elements must be identified. Several studies considered the release of toxic elements from coal combustion (3-7) or fossil-fuel combustion in general (8).Despite the large mass of material released from coal combustion, analyses by Gladney et al. ( 3 ) and Small (7) indicate that, aside from As and Se, coal cannot account for most toxic elements in urban particulate matter. Furthermore, few elements from coal-fired plants are predominantly associated with small, respirable particles, although some fractionation of volatile elements toward smaller particle sizes has been demonstrated (3, 6, 7). Present address, Analytical Chemistry Division, National Bureau of Standards, Washington, D.C. 20234. 566

Environmental Science & Technology

Little was known about the emissions of various elements from incinerators. This information is needed because of possible environmental effects of existing incinerators and because many communities are considering the use of refuse-derived fuel (RDF) for heat or electric power generation. Before plans for these systems become fixed, the possible release of toxic substances from refuse incineration must be understood. We have studied particles released by incineration of urban refuse in two municipal incinerators in the Washington, D.C., area: the Alexandria (Va.) Municipal Incinerator and the Solid Waste Reduction Center #1 in Washington, D.C.

Sites Studied The Alexandria incinerator has two identical furnace trains each capable of incinerating 140 metric tons of refuse daily. Refuse is fed into the primary combustion chamber where it passes over a series of rocking grates. The combustion gases and suspended particles pass into the secondary combustion

0013-936X/78/0912-0566$01 .OO/O

0 1978 American Chemical Society